Low noise BM800 mod

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
Yes, someone has, about a year before that video :)

http://audioimprov.com/AudioImprov/Mics/Entries/2015/12/20_Modding_a_BM-800_Mic.html

That being said, i'm not too crazy about 1) that add-on electrolytic having legs THAT long, and 2) it being tacked on to those flimsy little traces (and nothing else).
The least he could've done was to solder the legs onto the slighly-more-substantial  resistor pads nearby, along said traces.

PS: The PNP that's after the FET is drawn upside down in that 4chan schematic.

Watching through the video - hey, i think the guy had found the above link himself :p
"Didn't have a 470R resistor" - why not parallel two of'em? :D
 
Just wondering what is the real noise source for BM800 circuit. When playing with the circuit I removed the FET, replaced the following PNP with 2SK170 using 1G gate resistor and simple bias point for it (FET source  =>2.2k => bias point => 1.5k => GND).  Also removed the regulator transistor and added the extra filtering state.

It's much better now, but I noticed that when I switch off the phantom from my mixer used for audio tests (old Behringer) there isn't any improvement at first, but after about half a minute (discharging the phantom source capacitor) the noise level drops considerably.

So I'm wondering is there anything to do to drop noise in Schoeps type circuit if the noise source (the last few db's of it) is actually the output PNP's and it is somehow dependent on or related to the phantom voltage level? Dropping phantom voltage on mic pre could propably help, but it's not usually possible. Lower noise PNP's could be also used, but is the topology limiting the performance here in the end?
 
About the phantom voltage, the first phantom-powered mic, Schoeps CMT 20 in 1964, used 8.5V phantom voltage, later the CMT 200 accepted 9-12V phantom. Those were RF mics though, next year first FET-mics were announced, the CMT 30 F (using negative 12V phantom voltage) and later CMT 30 FN(and FNU, U meaning XLR connector) using positive 12V phantom.

Maybe it's best to build a dedicated mic pre with 12V phantom. That's  easy if you use THATs! THAT1512 for an example... No need for separate high voltage  phantom supply.
 
mhelin said:
Just wondering what is the real noise source for BM800 circuit.....
I found that the primary noise source was the zener diode, which - along with its associated regulator transistor  and lack of subsequent low pass filter - contributed most of the noise.
Unusual configuration - presumably  an attempt to allow these microphones to (sort of) work with computer sound card plug in power, as well as with phantom power?..

Certainly introducing an additional low pass filter can make a dramatic improvement to the noise floor.

But it will then only work with phantom power!

The capsule fitted into many BM800 mics is either a TSB 160A  - or very similar - clone. 
Not a bad capsule, but limited somewhat by being an electret, and a bit 'edgy' in the lower mid range, in my experience.

I think you can obtain much more dramatic improvements by simply retaining the body, connector and head basket of the BM800, and replacing the capsule with a 34mm capsule - some of which are now quite cheap.
Creating a simple Schoeps type circuit, along with  60 Volt capsule bias and you can get a really quite a good mic --- dependent mostly on the capsule quality!

I've built several of these, based on my notes here: http://www.jp137.com/lts/LDCX2.notes.pdf

Using low noise transistors, along with a simple (no inductors!) DC voltage multiplier  based around Rory Holmes 1983 ETI magazine article (see here:  http://www.gyraf.dk/schematics/Voltage_multipliers_with_CMOS_gates.pdf ) and you can get a pretty low noise - full bandwidth - mic for relatively little money.
Certainly the 6dB of sensitivity improvement over using a 30V (or so) DC bias is well worth the effort, IMHO..

(I think that type of voltage multiplier has been used by  other mic builders over the years?....)



 
Thanks, yes, the zener noise can be an issue. Doug Ford of Rode Mics recommends using soft knee zener for lower noise. The "knee" seems to be depending mostly on voltage, the lowest voltage zeners having softer knee. So connecting small voltage zeners in series is one way. Using LED's (blue ones have about 3.3V forward voltage) in series. Have to test. I kind of like the sound of the electret capsule but only when there is no noise. The grille affects also lot to the sound, maybe the BM700 would be a better choice. Anyway, the whole Schoeps circuit was originally designed to work at 12V voltage. Maybe there's not much to gain using that but anyway.
 
mhelin said:
The grille affects also lot to the sound, maybe the BM700 would be a better choice. ...
BM700 possibly, if it has the same double mesh head basket as the BM800 ?......
The other common cheap 'chisel shaped' head basket type - the NW700 - only comes with a single mesh head basket (the capsule there is a cheaper electret with an internal FET).
That type definitely needs a second mesh fitted if it is used with an external FET  type capsule.......At that impedance, it hums like hell without the second mesh! 
 
BM700 is similar to BM800 but has U87 style head basket, dual layer I think.
Experimented using couple of LEDs in series but it wasnt much better than zener. Only difference was that the zener was not emitting any light! Should really put those leds inside basket to light when phantom is on. Next modification will be the PNPs.
 
Is this using the stock board / circuitry?

And if you're referring to the EEVblog videos with Doug Ford, you'll recall him mentioning the solution to that hard-knee zener problem - "Whack an RC filter after it" ;D
 
Stock board, just trying to explore the circuit, nothing serious. Event though I didn't watch the whole Eevblog video I can remember that sentence. Zener stuff starting at 14:30 on this video:
https://www.youtube.com/watch?v=WTJhIVIGvSU&list=PLvOlSehNtuHv98KUcud260yJBRQngBKiw&index=4

At 21:30 he is drawing the Rode Classic II circuit. I think that the mic actually uses PNP and not NPN, Doug mentions 2N5401 which is also PNP. There are some other errors too, in the Phantom Powering part he draws Schoeps circuit but I think it goes south at some point.

That link above btw. takes to the Microphone Technology and Design series and lists all of those Doug Ford videos.

 
analogguru said:
A simple 470 ohm resistor between the zener-diode and base of the regulator transistor (together with the filter cap) will work wonders.

Exactly what I did in the suggestion I referred to in my reply #4.  I also  added a similar low pass filter to my  own circuit (shown in my notes I linked to  in  that same reply.)

I had originally used  an obselete 2N5457 FET I had lying around in my 'schoeps' variation , but changed that for a more easily obtainable J113 ( largely similar characteristcs )
I selected components to bias that to allow about 1mA of current to flow  through the FET.  That seems to work very well.

That meant that I could afford to lose about 0.5V  through the 470R low pass filter resistor, and still retain a useful range of bias voltage  for the  preamp.  I decoupled the output side of the 470R with both 220uF and 100nF capacitors. The electrolytic does most of the work, with the 100nF 'catching' any HF noise that eludes the bigger cap.

I selected 9.1V zener as a value to allow  the DC multiplier to generate 60 Volts. Going higher would have meant using  some 100V capacitors (rather than 63V), and the sensitivity advantage of going up to - say - 70V is probably not worth the hassle?...

What did surprise me was how well the circuit performs with a (not very good) 24V phantom supply. 
One of my old Behringer mixers supplies  24 V phantom - (well 'sort of' - it's actually around 20V off load!)

That still gave me around 6.8V to power the preamp and the multiplier... so the zener wasn't even conducting!
Around 45 V capsule bias, with no significant increase in noise..
Seems to work pretty well for a such a simple circuit......

 
analogguru said:
A simple 470 ohm resistor between the zener-diode and base of the regulator transistor (together with the filter cap) will work wonders.

Ok, so you have the NPN collector => 22k =>  zener (to GND) => 470 ohms (only new component)  => 47 uF (to GND) =>  NPN base and the output from the NPN emitter. Isn't this circuit also a capacitance multiplier (should we use a high gain transistor)?

Have several of those BM800's, got to test just the above and compare the results.
 
mhelin said:
Ok, so you have the NPN collector => 22k =>  zener (to GND) => 470 ohms (only new component)  => 47 uF (to GND) =>  NPN base and the output from the NPN emitter. Isn't this circuit also a capacitance multiplier (should we use a high gain transistor)?
Exactly - and you will not loose 0,5 V across the resistor due to the low base current (It will produce a clean output even when the zener voltage would not be reached).

mhelin said:
Have several of those BM800's, got to test just the above and compare the results.
Verifiy that the zener-diode is 9,1V.  On some units it is only 3,1V because someone thought it could be powered from the (5V-)PC-mic-input too.  Forget about this, in most cases it will not work and people have to buy additionally a phantom power supply.  No-one needs a schoeps circuit for a PC-mic-input.

And yes, sometimes the voltage divider at the base of the HP-Filter/Phase-splitter-transistor is wrong.  For best performance (max headroom) the voltage across the collector- or emitter-resistor should be approx a quarter (~ 2,2V) of the supplying voltage (~ 9V).

rogs said:
Exactly what I did in the suggestion I referred to in my reply #4.  I also  added a similar low pass filter to my  own circuit (shown in my notes I linked to  in  that same reply.).....
Sorry, I couldn't find a (voltage) regulation transistor in your notes like it can be found in the original BM-700/BM-800 circuits.  I was talking about a simple mod using the original pcb.
 
analogguru said:
Sorry, I couldn't find a (voltage) regulation transistor in your notes like it can be found in the original BM-700/BM-800 circuits.  I was talking about a simple mod using the original pcb.

Sorry , my bad explanation!... I think the series transistor on the BM800 is only there to help this slightly unusual configuration work with very low DC levels.
Some folk say it will work with just the bias from a computer sound card.... Others say not? .... It certainly works better with 'real' phantom power !

I had tried out the low pass filter mod on a BM800 and found it quite successful, so I thought I would add it to my 'schoeps' circuit as well.
I realised that the original Schoeps schematic had only de-coupled the zener with an electrolytic, and didn't include a second low pass filter, but as it had worked well on the BM800 I thought I'd include on my new schematic.
It only added a few pennies to the cost  and  - as I mentioned - only lost me about 0.5V of DC potential to the FET bias chain.
The extra 100nF 'HF' de-coupling cap is probably a bit over the top?.. but again, I have found improvements by using it in other circuits, and decided it was worth the extra 5p in cost!

The combination of using 60V to bias the LDC capsule, and the use of low noise PNP transistors in my simple circuit, has given me a mic with a noise floor over 20dB better than the original BM800 -- and only some 8 or 9dB worse than my (very!) low noise Rode NT1...

 
 
I believe i recall Doug Ford mentioning, somewhere in the above-linked video, about capacitors being sort of almost useless for "de-noising" zeners, since the zeners will always(?) have lower impedance than the capacitor.

So i can see why a resistor between the zener and the capacitor would be more effective (ie. not having a capacitor in parallel with the zener at all).

rogs said:
I realised that the original Schoeps schematic had only de-coupled the zener with an electrolytic, and didn't include a second low pass filter, but as it had worked well on the BM800 I thought I'd include on my new schematic.
 
rogs said:
Sorry , my bad explanation!... I think the series transistor on the BM800 is only there to help this slightly unusual configuration work with very low DC levels. .....
No, the transistor (together with the zener-diode) works as a voltage regulator.  This was a pretty standard circuit before you could buy integrated volage regulators.

Khron said:
.....
So i can see why a resistor between the zener and the capacitor would be more effective (ie. not having a capacitor in parallel with the zener at all).
Another benefit of this resistor is that it acts as a protection for the zener-diode in case when the transistor is damaged (think of a C-B-short) even if this will not happen in this circuit.



 
analogguru said:
No, the transistor (together with the zener-diode) works as a voltage regulator.  This was a pretty standard circuit before you could buy integrated volage regulators.

I can see that the transistor will work in conjunction with the zener  as a voltage regulator.
I just can't see that in any 'normal' Schoeps style configuration you would need to include a transistor as well as a zener? .... we're not talking about drawing any serious current here....
And doesn't that series transistor amplify  the zener noise?..
I can only assume it is configured like that for use with DC supplies that  don't even reach the zener voltage. Then the 22k resistor between base and collector will serve to keep the transistor turned on and minimise the impedance of what ever low voltage DC supply is  'trying' to supply the circuitry.
The cables that come with these mics are unbalanced XLR to 3.5mm mono jack sockets, so they're obviously expecting some customers to try an unbalanced connection to a computer card input.
As I mentioned before, some folk seem to get it working like that - others not.




 
Back
Top